{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:OB6QXQEW7RJ3UYOR5MRV5KP6JZ","short_pith_number":"pith:OB6QXQEW","schema_version":"1.0","canonical_sha256":"707d0bc096fc53ba61d1eb235ea9fe4e7febbfa0ea194446b98a1455cd0d722f","source":{"kind":"arxiv","id":"2607.17823","version":1},"attestation_state":"computed","paper":{"title":"Theoretical Foundations of $\\max$@$k$ Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Andrea Celli, Martino Bernasconi, Riccardo Poiani","submitted_at":"2026-07-20T11:11:55Z","abstract_excerpt":"Reinforcement Learning is a cornerstone technique for modern large reasoning models. Usually, for difficult tasks such as code generation and theorem proving, the agent is evaluated by generating $K$ responses rather than sampling a single response, and performance is then measured using a retry-aware metric such as $\\max$@$k$. Despite their practical importance, the theoretical foundations of learning under such criteria remain limited. In this work, we provide a theoretical study of the $\\max$@$k$ learning problem in finite-horizon reinforcement learning. We show that optimizing the $\\max$@$"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2607.17823","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-07-20T11:11:55Z","cross_cats_sorted":[],"title_canon_sha256":"68f92077a7114653fdc1188e9cd96252179e9009d99d6543ab4157a0de65f140","abstract_canon_sha256":"863fffa505966f87b7385c9586f149032093a8325b4e26011249a06c60f48c42"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-21T02:22:02.099088Z","signature_b64":"16qavS1naDAcfqLG4ueNCR90YUWxdfnQZFOviPDQIHUE5XmHyWtPAIuhcl64rhWmxMku+8pmeN/I1WJ6stSmBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"707d0bc096fc53ba61d1eb235ea9fe4e7febbfa0ea194446b98a1455cd0d722f","last_reissued_at":"2026-07-21T02:22:02.098277Z","signature_status":"signed_v1","first_computed_at":"2026-07-21T02:22:02.098277Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Theoretical Foundations of $\\max$@$k$ Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Andrea Celli, Martino Bernasconi, Riccardo Poiani","submitted_at":"2026-07-20T11:11:55Z","abstract_excerpt":"Reinforcement Learning is a cornerstone technique for modern large reasoning models. Usually, for difficult tasks such as code generation and theorem proving, the agent is evaluated by generating $K$ responses rather than sampling a single response, and performance is then measured using a retry-aware metric such as $\\max$@$k$. Despite their practical importance, the theoretical foundations of learning under such criteria remain limited. In this work, we provide a theoretical study of the $\\max$@$k$ learning problem in finite-horizon reinforcement learning. We show that optimizing the $\\max$@$"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.17823","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2607.17823/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2607.17823","created_at":"2026-07-21T02:22:02.098694+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.17823v1","created_at":"2026-07-21T02:22:02.098694+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.17823","created_at":"2026-07-21T02:22:02.098694+00:00"},{"alias_kind":"pith_short_12","alias_value":"OB6QXQEW7RJ3","created_at":"2026-07-21T02:22:02.098694+00:00"},{"alias_kind":"pith_short_16","alias_value":"OB6QXQEW7RJ3UYOR","created_at":"2026-07-21T02:22:02.098694+00:00"},{"alias_kind":"pith_short_8","alias_value":"OB6QXQEW","created_at":"2026-07-21T02:22:02.098694+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OB6QXQEW7RJ3UYOR5MRV5KP6JZ","json":"https://pith.science/pith/OB6QXQEW7RJ3UYOR5MRV5KP6JZ.json","graph_json":"https://pith.science/api/pith-number/OB6QXQEW7RJ3UYOR5MRV5KP6JZ/graph.json","events_json":"https://pith.science/api/pith-number/OB6QXQEW7RJ3UYOR5MRV5KP6JZ/events.json","paper":"https://pith.science/paper/OB6QXQEW"},"agent_actions":{"view_html":"https://pith.science/pith/OB6QXQEW7RJ3UYOR5MRV5KP6JZ","download_json":"https://pith.science/pith/OB6QXQEW7RJ3UYOR5MRV5KP6JZ.json","view_paper":"https://pith.science/paper/OB6QXQEW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.17823&json=true","fetch_graph":"https://pith.science/api/pith-number/OB6QXQEW7RJ3UYOR5MRV5KP6JZ/graph.json","fetch_events":"https://pith.science/api/pith-number/OB6QXQEW7RJ3UYOR5MRV5KP6JZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OB6QXQEW7RJ3UYOR5MRV5KP6JZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OB6QXQEW7RJ3UYOR5MRV5KP6JZ/action/storage_attestation","attest_author":"https://pith.science/pith/OB6QXQEW7RJ3UYOR5MRV5KP6JZ/action/author_attestation","sign_citation":"https://pith.science/pith/OB6QXQEW7RJ3UYOR5MRV5KP6JZ/action/citation_signature","submit_replication":"https://pith.science/pith/OB6QXQEW7RJ3UYOR5MRV5KP6JZ/action/replication_record"}},"created_at":"2026-07-21T02:22:02.098694+00:00","updated_at":"2026-07-21T02:22:02.098694+00:00"}